Coding Better Composables: Start with the Interface (4/5)

The composable you wrote last month returns an object, but right now, you really need it to return an array instead. But it’s too late to change now. That would require literally days of refactoring since the composable is already all over the codebase. If only you had taken a bit more time, in the beginning, to really figure out how you’d want to use the composable, you might have been able to save yourself from this frustration…

In this article, we will learn about writing interfaces for our composables and how this process can help us craft composables that serve us far into the future. This article is the fourth in a five-part series on coding better composables. If you missed the first three articles, you’ll want to start from the beginning. This series will walk you through several best practices when writing composables. Once you’re through, you’ll clearly understand crafting solid composables.

Now without further adieu, let’s get into the Interface.


Write the Interface First

Instead of starting with the implementation, you should start with how you want the composable to be used. That’s what I mean when I say, “begin with the interface.” It’s the API of the composable. It’s the boundary between the composable and other pieces of your app.

It’s tempting to create a new file and start coding away on the implementation, but it’s rarely the best approach. Instead, it’s easier to prototype when you start with the interface, and you’re also likely to create a better interface.

This is important because the interface is way harder to change 6 months from now than the implementation is. So it’s better to spend extra time on this step getting it right.

This is especially true when dealing with highly reusable code. Most reusable pieces of code, whether a component or composable, undergo a lot of volatility when they’re first created. The interface keeps changing because we keep discovering edge cases that require us to refactor. Instead, if we can anticipate most of those edge cases initially, we can save ourselves a lot of pain and frustration.

(But unfortunately, we can never predict the future 😢)

To start with our interface, we write our code pretending that the composable already exists. We also have to answer a few questions through this process:

  1. What arguments do we pass into our composable? A ref, a raw value, or a series of values?
  2. What options should be included in the options object?
  3. What values does our composable return? Is it just a single value, or do we want to use the dynamic return value pattern?

These don’t have to be answered in a specific order, but most should be figured out. Also, I find that I do this iteratively. As I start to answer one question, I better understand how to answer the others. These questions serve as more of a guide than a rigid process you need to follow.

You’ll notice each of these questions is only concerned with how we interact with and use the composable. We’re not yet worried about how the composable actually works. Of course, we’ll get to that part, but only after we’ve figured out the interface.

So why don’t we work through an example to see this process in action?


Figuring out the useMouse interface

We’ll go through a bit of a sketch to see how this can play out, with a useMouse composable. This composable will give us access to the mouse coordinates on the screen. Let’s answer each question and see how we can figure out this interface.

Question 1: What arguments do we pass into our composable?

The answer to this question is straightforward — the useMouse composable doesn’t take in any arguments. Instead, we want the mouse coordinates to return to us and update reactively.

Question 2: What options should be included in the options object?

Since the composable will be updating the mouse movements, it might be nice to have control over how often those updates happen. For this, we can have an interval option to specify that timing.

When it comes to on-screen coordinates, an excellent question is: relative to what? With mouse coordinates, we can get them relative to:

  • The browser’s viewport
  • The entire document
  • A specific element

So we should have a second option — target — that lets us choose what we’re basing our coordinates on. A good default value would probably be the browser viewport.

Question 3: What values does the composable return?

This one is not so clear from the start. Sometimes we just need to start writing code that uses the composable, and along the way we discover how we want to use it. That process can give us a lot of clues as to what our interface should be.

Maybe we start by grabbing the coordinates as an object:

const { x, y } = useMouse();

As we continue prototyping, we realize we also need to know if the mouse is being clicked or not, so we’ll add that in:

const { x, y, isClicked } = useMouse();

This works, but it’s kind of clunky. As we continue writing our app, we realize that what we really need is to grab the mouse coordinates in a separate object:

const { position, isClicked } = useMouse();

Perhaps, as we continue building this out, we discover that sometimes it’s handy to have the coordinates as an array. We can include that in our interface:

const {
  position,
  positionArr,
  isClicked,
} = useMouse();

At this point, we have a pretty good understanding of how we’ll want to use the useMouse composable. But, of course, requirements change over time, so this isn’t set in stone. But you can see how much our interface has changed by prototyping a bit. If we had just gone with our first guess, we might have written a composable that would need to be refactored soon.


Wrapping Things Up

In this article, we learned that by thinking critically about how we want to use our composable in the beginning, we can write a better implementation. And by writing a better implementation, we save ourselves time and frustration in the future. We won’t need to refactor as much because we’ve been thoughtful about how we want to use the composable.

Of course, we don’t want to spend too much time on this process. There is only so much that you can anticipate, and we all know that requirements can change often. Also, there is no perfect interface, but we can avoid many headaches with a bit of effort at the start.

The next and final article in this series will cover how we can simplify async code in our composables, preventing a bunch of bugs and confusing situations:

const count = ref(0);
// This async data fetch won't interfere with our reactivity
const { state } = useAsyncState(fetchData());
const doubleCount = computed(() => count * 2);

With this pattern, we don’t need to use await, and we can put async code wherever we want without worrying about it affecting reactivity.

Download the cheatsheets

Save time and energy with our cheat sheets.